A passenger's choice model of train service

The paper presents a modeling framework for solving the service planning problem for Taiwan high speed rail. The framework explicitly considers both the passenger's and the operator's viewpoints in a bi-level program, where a train demand model is the lower level problem. The passenger's choice of train service is formulated as a route choice problem on a service network and with a generalized cost function. In vehicle travel time, out of vehicle travel time, out of pocket cost, and crowding condition are the variables in the cost function. Three models, which are deterministic user equilibrium, stochastic user equilibrium, and deterministic user optimal with crowding effect, are tested in the study with practical data. The three models result in quite different choice patterns. The comparison of the estimated choice patterns can explain the effect of the modeling assumptions dearly. Moreover, the paper presents the sensitivity results of the time value, fare structure, train frequency, and so on.